Abstract:

MOOC format is characterized by the great diversity of enrolled people, which are shaped by different personal and professional backgrounds, many knowledge levels, dissimilar motivations and degrees of satisfaction, among many other features. All of them make up the profiles of the participants, whose lack of knowledge constitutes an important barrier in order to identify and get a better understanding of their underlying relationships.

This paper, whose scope is the data collection of MOOCKnowledge project, aims to identify and analyze the types of profiles of participants with previous experience in MOOC format according to their degree of prior satisfaction in those courses by running clustering techniques. These approaches fit the final purpose of finding groups of satisfaction (groups of participants) according to the similarity of the selected items for measuring their degree of satisfaction. The set of items through which is built up a participant's degree of prior satisfaction are, among others, course materials, video lectures, teacher, learning activities, assignments, feedback given, pace of the course, language used, MOOC difficulty, technical requirements and certification options.

Clustering approach is performed in this study with the set of items that shapes the prior satisfaction of participants in MOOC format as input data. The resulting groups of satisfaction, by running unsupervised clustering algorithms, are expected to have great similarity within each group but little similarity (dissimilarity) across groups.

The approach of this paper is focused on the similarities within the identified groups of satisfaction (types of profiles of participants' MOOC) as result of clustering process. The identification and further analysis of the hidden relationships in the internal structure of the set of items on each group might help stakeholders to identify the most relevant items that impact in a decisive way on participant's prior satisfaction regarding MOOC format.

Learner satisfaction should be a measure of success for any course and, based on this assumption, this study takes advantage of the exploratory nature of clustering techniques for identifying and analyzing participants' profiles from a set of items related to their degree of prior satisfaction with MOOC format.

It should be highlighted that this study, fed with a dataset that comes from MOOCKnowledge data collection, might help in setting up short and medium term strategies aimed not only to discover an individualized knowledge regarding the more influential items for the groups of satisfaction but also to analyze the changes in the degree of satisfaction of participants after the completion of the course.

Keywords:

MOOC profiles, educational data mining, clustering techniques.